TestBench

Tertiary structure Tertran Terylene Teslac TestBench Testing... 

K. Oblander, J. Abthoff, and L. Fricker, "From Engine Testbench to Vehicle—An Approach to Lean Bum by Dual Ignition," I. Mech E. CSOj79, (1979). 

In the case of our testbench, getint has all the integrals and their labels stored as part of itself and passes them out on demand. The full listing of this very specialised getint is given in Appendix 5.C to this chapter, the essence of it is that the labels and values of the repulsion integrals are stored in... 

The testbench in the last section was an implementation of conventional SCF the one-electron and repulsion integrals were computed once and stored for use during the iterations of the SCF procedure. It is just as easy to set up a testbench to demonstrate the other extreme the calculation of the energy integrals as they are required during the SCF iterations the so-called direct SCF method. 

The principle idea behind these testbenches is that the various special cases used to display the capabilities of the code developed earlier should be, as far as is reasonable, invisible to the design of the overall system no major restructuring should be necessary to the basic code. In the case of the electron-repulsion integrals, the function getint is changed from case to case and no other changes are needed. 

The remaining data are just the matrices of one-electron integrals and a matrix which will generate an orthogonal basis from the minimal beisis of AOs . Here are the data statements which supply these data to the testbench program ... 

The two testbench programs of the last chapter have a great deal in common they are functionally identical and only differ in a few details which depend on the use to which the code is to be put. There is an important general rule here which gives the opportunity of demonstrating the value of conditional compilation. 

One of several related programs which perform closely related tasks like the two related testbenches of the last chapter. 

The main difference between the two testbenches is the fact that the second of the two uses an orthogonal basis and so some of the steps in the SCF method are omitted. The relevant program fragment in the full calculation is... 

Of course, our two testbenches are merely that, examples to be discarded as experience is gained with the implementation, and the use of conditional compilation here is rather out-of-place but it docs provide a convenient introduction to the technique and its use. 

Once the initial oscillations have been damped, the interpolation procedure can be turned off since it may become a hindrance to convergence again at a point determined by experience. Putting this simple algorithm into the testbench... 

In deciding that the only practical way to compute the molecular orbitals of an arbitrary polyclectronic system was to expand these MOs as a linear combination of some basis functions we, did not give any thought to the form which these all-important functions might take. The testbench programs used to test the codes do contain some implicit assumptions on this point, but the point has not so far received any attention. We must now give the matter some consideration if we are to develop methods of calculation suitable for any molecule, radical or ion. 

These considerations would seem to point unequivocally to the use of AOs as the natural expansion functions for MOs and, indeed, this is the assumption implicit in the two testbench programs. 

The Hartree-Fock equations which arise from the application of the variation principle to any of these constrained models guarantees a minimum in the energy of the associated single-determinant wavefimction. What they do not guarantee is that this wavefunction is physically meaningful. This point wiU be taken up in detail later for the moment simply recall the single-determinant solution of the closed-shell model for (H20) obtained with the testbench which had positive orbital energies ... 

If the basis size is doubled, the whole of the testbench program can be used as it stands for a GUHF program if we write a new scfGR procedure and arrange for the orthogonalisation and one-electron Hamiltonian matrices to be doubled and stored correctly. 

We simply have to insert data statements with the values of the elements of the matrices h and V into our closed-shell SCF program to have a working testbench. These data are also relegated to Appendix 5.C. 

This last point can be made even more strongly by proceeding to the unphysical limit of the basis. A calculation on (H20), obtained by replacing n = 5 by n = 7 in the testbench code generates the following output ... 

---